In 1-6 experiment_{index}/ folder, you will find the following files:
1. one_step_training_ex_{index}.ipynb: A Jupyter Notebook that contains the code used to train the model for the corresponding experiment. It includes all the steps for data loading, model training, and saving the best weights.
1.`one_step_training_ex_{index}.ipynb`: A Jupyter Notebook that contains the code used to train the model for the corresponding experiment. It includes all the steps for data loading, model training, and saving the best weights.
2. one_step_evaluation_ex_{index}.ipynb: A Jupyter Notebook used to evaluate the performance of the trained model. You can run this notebook with the evaluation results (local_evaluation_ex_{index} and test_set_evaluation_ex_{index}) obtained from the training notebook to compute the mAP for both the local and test sets.
2.`one_step_evaluation_ex_{index}.ipynb`: A Jupyter Notebook used to evaluate the performance of the trained model. You can run this notebook with the evaluation results (local_evaluation_ex_{index} and test_set_evaluation_ex_{index}) obtained from the training notebook to compute the mAP for both the local and test sets.
3. test_set_evaluation_visualization_ex_{index}.ipynb: A Jupyter Notebook used to visualize the performance of the trained model. You can run this notebook with the evaluation results (test_set_evaluation_ex_{index}) obtained from the training notebook to visualize test images with bounding boxes.
3.`test_set_evaluation_visualization_ex_{index}.ipynb`: A Jupyter Notebook used to visualize the performance of the trained model. You can run this notebook with the evaluation results (test_set_evaluation_ex_{index}) obtained from the training notebook to visualize test images with bounding boxes.
4. best_weights_ex_{index}.params: This file contains the best model weights that were obtained during the training of the corresponding experiment.
4.`best_weights_ex_{index}.params`: This file contains the best model weights that were obtained during the training of the corresponding experiment.
5. local_evaluation_ex_{index}.txt: The val output of training notebook.
5.`local_evaluation_ex_{index}.txt`: The val output of training notebook.
6. test_set_evaluation_ex_{index}.txt: The test set output of training notebook.
6.`test_set_evaluation_ex_{index}.txt`: The test set output of training notebook.
7. test_set_evaluation_ex_{index}.png: Screenshot of the corresponding experiment results, which include output from Hugging Face.
7.`test_set_evaluation_ex_{index}.png`: Screenshot of the corresponding experiment results, which include output from Hugging Face.
In experiment_7 Folder, the process is split into two stages: detection and classification. The workflow includes the following key files:
1. detection_training_ex_7.ipynb: Contains the code for training the detection model, which detects fruit regions without classifying them.
2. detector_best_weights_ex_7.params: Stores the best weights for the detection model after training.
1.`detection_training_ex_7.ipynb`: Contains the code for training the detection model, which detects fruit regions without classifying them.
3. classification_training_ex_7.ipynb: Contains the code for training the classification model, which classifies the detected regions.
2.`detector_best_weights_ex_7.params`: Stores the best weights for the detection model after training.
4. classifier_best_weights_ex_7.params: Stores the best weights for the classification model after training.
3.`classification_training_ex_7.ipynb`: Contains the code for training the classification model, which classifies the detected regions.
5. two_step_inference_ex_7.ipynb: Runs inference by first detecting fruit regions and then classifying them, producing results for both detection and classification.
4.`classifier_best_weights_ex_7.params`: Stores the best weights for the classification model after training.
Additional files include detection and classification results for both the test and validation sets, as well as cropped images generated during detection.
5.`two_step_inference_ex_7.ipynb`: Runs inference by first detecting fruit regions and then classifying them, producing results for both detection and classification.
Additional files include detection and classification results for both the test and validation sets, as well as cropped images generated during detection.